<p>The flow behavior and hot workability of 022Cr19Ni10 austenitic stainless steel at elevated temperatures were investigated to enhance its performance in the spent nuclear fuel reprocessing environment. A single-pass compression test was conducted on 022Cr19Ni10 austenitic stainless steel using the Thermecmaster-200KN thermal simulation testing machine. Flow stress curves were obtained under deformation temperatures ranging from 900 to 1200&#xa0;°C, with strain rates varying within 0.01-10&#xa0;s¹, and a true strain of 0.7. The impacts of adiabatic and friction effects on the flow stress were comprehensively evaluated, and the curves were corrected accordingly. Based on the corrected flow stress data, the strain-compensated Arrhenius constitutive model and machine learning models such as BPNN, PSO-BPNN, and GA-BPNN were constructed and their predictive capabilities for the flow behavior were compared. The accuracy of the GA-BPNN model was further verified via DEFORM-3D. The hot workability of the material was analyzed through a processing map. The results show that friction leads to the measured flow stress exceeding the theoretical values, and the adiabatic effect intensifies with the increase in strain and strain rate, causing a transition from dynamic softening to dynamic recovery at low temperatures and high strain rates. The strain-compensated Arrhenius model exhibits poor fitting accuracy under material instability conditions, while the GA-BPNN machine learning model accurately captures the flow stress throughout the entire hot deformation process. The average relative error between the GA-BPNN predicted flow stress and the corrected values is merely 0.13%, and the simulation results are highly consistent with the experimental displacement-load curves. The processing map determines the optimal processing windows for a strain of 0.7 as 960-1100&#xa0;°C at strain rates of 0.01-0.025&#xa0;s⁻¹, 1050-1200&#xa0;°C at 3-8.5&#xa0;s¹, and 1100-1200&#xa0;°C at 0.01-0.4&#xa0;s¹.</p>